Phase 8 — Synthetic-Layer Assembly Report

Date: 2026-04-27 Run: First Phase 8 run; Pass 4 of v0c migration. Operational mode: Claude Code (executable plan; R8 supported-promotion rule = halt). Inputs: v0c-migration-kickoff.md; wiki/.audit/userMemory-snapshot-2026-04-27.md (especially the Pass 4 preliminary-triage handoff section); AUDIT_PLAN.md §Phase 8 + R8; CLAUDE.md §Claims Register Format and §Sourcing Rules for Claims; wiki/log.md recent entries (Pass 1, Pass 2, Pass 3a, Pass 3b commits). Cap (per Step 8.6): ~10–15 live, ~3–5 supported, ~10–15 concept-page upgrades. Pass 3 already discharged the concept-page upgrade quota in Passes 3a and 3b (3 new + 7 in-place = 10). Pass 4's substantive work is claims.md population; concept-page touches are the cite-back walk in Pass 5 plus the Heinbokel false-friend caution.

Headline numbers

  • Candidates surveyed: 11 kickoff seed claims + 1 false-friend kickoff entry + the empirical-confirmation user-memory observation = 13 considered nominees.
  • Promoted to live: 5 (Cand 1 reformulated, Cand 7, Cand 9, Cand 10+12 merged, Cand 11).
  • Promoted to supported: 0 (per R8: no candidate plainly meets the 5-test gate on first walk; conservative first-run target was 0 supported promotions).
  • Stayed at candidate with documented evidence gap: 5 (Cand 2, Cand 3, Cand 4, Cand 5, Cand 6).
  • False-friend cautions added: 1 (Heinbokel medical-phenomenology register on [[coherent-deformation]]).
  • Non-claim recorded for audit-record completeness: 1 (the empirical-confirmation observation about the LLM ingester failing to extract science secrète — kept as kickoff-recorded "page citing itself" caution; not entered in claims.md).
  • Duplicate-merge decisions: 1 (Cand 10 and Cand 12 merged into a single entry per kickoff, with the universalization thesis as the main claim and the symbolic-matrix-as-instance as a corollary sub-thesis).
  • Supported-promotion confirmations filed: 0 (no candidate reached the supported gate; no wiki/.audit/phase-8-supported-confirmations-2026-04-27.md filed).

The first-run claims register has 10 entries (5 live + 5 candidate). All within the kickoff cap.

Walk order followed

Per AUDIT_PLAN Step 8.2:

  1. Extraction-note Diagnostics + Recurring Motifs sections — surveyed via the kickoff seed list as the high-yield filter. The 11 seed candidates already encoded the kickoff's reading of the extraction-note Diagnostics, and the prior triage in userMemory-snapshot-2026-04-27.md had pre-classified each. This walk validated the triage rather than re-deriving it.
  2. Source-page "What's Not Obvious" sections — partially exercised through the cross-source-motif arguments captured in the kickoff seed list. Not separately walked end-to-end this run; the 5 live promotions already saturate first-run capacity. Future audits should cycle through the WNO sections as a first-class walk-order step.
  3. wiki/motifs.md BRIDGE entries — not walked end-to-end this run. Per snapshot triage Phase 8 walk-order remaining (item 3): the 7 cross-tradition BRIDGE entries — asymmetric bi-directionality (Schelling↔MP, already operationalized on chiasm.md / dass-was-distinction.md per 2026-04-23); thinking-away/unvordenklich ↔ incorporation-of-truth; self-falsification ↔ chiasm; circulus-vitiosus-deus ↔ V&I; granite-of-fate; Stiftung Husserl↔MP (now covered by stiftung.md); Bergson↔MP↔Deleuze — may yield 1–3 additional structural-parallel claim entries in subsequent runs. Cap-bounded deferral, not omission.
  4. Prior audit reports in wiki/.audit/ — applied the Rule 17 audit-report-authorization filter:
    • paper-a-gaps-2026-04-25.md (Phase 1.4 gap report) — cites traceable anchors; qualifies as evidence anchor for genealogical / problem-space claims about Paper A's gaps. Used implicitly through the kickoff's seed candidates.
    • silent-keys-2026-04-25.md (Phase 2 silent-key report) — cites traceable anchors; the science secrète / équivalent interne / Klee fundamental-indirect-absolute painting findings anchor claims#science-secrete-stiftung-chiasm Evidence.
    • saint-aubert-backlog.md (Phase 4 backlog) — cites traceably-anchored material; qualifies. Not directly load-bearing for this run's promotions.
    • citation-integrity-2026-04-25.md (Phase 6 spot-check) — procedural, not claim-authorizing. The fabricated-Faul-citation note in this report informs Cand 3 / Cand 5 Counterpressure framing.
  5. User-memory thesis nominations — userMemory-snapshot-2026-04-27.md is the load-bearing nominative artifact. Per Rule 17 it cannot authorize promotion above candidate; the artifact-anchor verifications happened against extraction notes, not against the snapshot.
  6. Seed claim candidates — the 11 kickoff seeds, walked in triage order. Each promotion / candidate decision is recorded under the corresponding claims.md entry's Status History.

Triage decisions per kickoff seed candidate

The pre-walk triage in userMemory-snapshot-2026-04-27.md classified each candidate; this walk validated each against the extraction notes named in the kickoff. Reproduced here with disposition.

Seed # Disposition Slug Notes
1 live (reformulated) science-secrete-stiftung-chiasm E&M extraction Silent Key #1 (line 341) and V&I extraction Pass 2c (line 300) anchor the joint-operation thesis. Per Pass 1 commit B2: Saint Aubert moved to Counterpressure; depends on E&M + V&I only.
2 candidate nov-1960-stiftung-grammatical-subject V&I extraction line 300 records the working note but the philological grammatical operative subject reading needs targeted raw check of Fr 320–321 per Rule 18. Stays candidate.
3 candidate cryptic-institution-extends-beith Faul not in raw/. Stays candidate with explicit Faul-ingest precondition.
4 candidate valery-mauvaises-pensees-attribution Valéry Œuvres II IS in raw/ per AUDIT_PLAN v1.2 erratum B1, but has not been ingested (no extraction note, no source page). Malraux Voices of Silence not in raw/. Stays candidate with explicit dual-ingest precondition.
5 candidate faul-institutional-camp Faul not in raw/. Stays candidate.
6 candidate deformation-coherente-mp-coinage Johnson 2020 extraction note records the concept reference but not the attribution thesis; per Rule 17, user-memory cannot authorize. Targeted Johnson re-read deferred per 2026-04-25 audit log line 804. Stays candidate.
7 live ingested-corpus-four-element-gap Negative-evidence claim, scoped to ingested secondary corpus only. 16-source enumeration computed 2026-04-27; zero attestations of science secrète across all secondary sources, only 3 with coherent deformation, none combining all four elements. Counterpressure documents corpus boundary, not-in-raw/ gaps (Saint Aubert 2006, Faul, Andén, Fóti, Kee), and keyword-method limit.
8 not a claim — false-friend caution (n/a) Heinbokel medical-phenomenology register added as Open Questions / False-Friend subsection on [[coherent-deformation]] this run.
9 live kaushik-stiftung-literary-frame Kaushik 2019 extraction note (lines 7, 60, 75) and Kaushik 2021 extraction note (lines 66, 117, 144) anchor the literary-language framing; the painter's-body / science secrète / indirect-ontology gap is documented. Counterpressure: gap is primary-framing, not topical absence.
10+12 (merged) live ineinander-universalizes-institution Merged per kickoff. V&I extraction line 300/334/584 + I&P extraction line 88/133/143 + PoP extraction App-A2/A8 + Kaushik 2019 line 75 anchor the universalization thesis with symbolic-matrix-as-instance corollary.
11 live ip-pop-architectural-hierarchy I&P extraction line 84/143/175 + PoP extraction C1-A12/A14 + App-A1/A2 anchor the architectural hierarchy. Counterpressure: dominant secondary reading sees late MP as moving away from institution. PoP "p. 79 Heidegger malaise" raw-passage detail not specifically anchored — flagged in Counterpressure as targeted raw check that would tighten the chain (not blocking; the architectural-hierarchy claim is supportable from the I&P fold-into framing plus PoP App-A1/A2).

Broken evidence chains queued for targeted raw checks / future ingests

The 5 candidate-only entries each name an evidence chain that depends on a specific anchor not yet available in the wiki. These chains form the explicit Phase 8 backlog — promotion to live requires the corresponding artifact:

  1. Cand 2 (nov-1960-stiftung-grammatical-subject) — targeted raw-source check of V&I Fr 320–321 (in MP's Le Visible et l'invisible, Gallimard 1964). Per Rule 18 (artifact conservatism): passage-level local-context verification, not full re-ingest. The V&I source IS in the wiki and presumably in raw/ — the gap is a re-read of one passage, not a missing source.
  2. Cand 3 + Cand 5 (cryptic-institution-extends-beith, faul-institutional-camp) — Faul, "Ontologically Interactive Painting" not in raw/. Both candidates require Faul ingest. Note: any reference to Faul must use the real paper title; the fabricated "The Painter's Courage" / JBSP 55:2 / 2024 citation has been removed (see AUDIT_PLAN Phase 6 spot-check).
  3. Cand 4 (valery-mauvaises-pensees-attribution) — Valéry Œuvres II Pléiade source IS in raw/ (per AUDIT_PLAN v1.2 erratum B1: OEuvres (Pleiade), tome II - Paul Valery.md) but has not been ingested. Malraux Voices of Silence is not in raw/. Promotion requires both ingests.
  4. Cand 6 (deformation-coherente-mp-coinage) — targeted re-read of Johnson's essay in johnson-carbone-saintaubert-2020-poetic-of-the-world to locate the specific attribution argument. The Johnson 2020 extraction note exists but doesn't anchor this thesis. Deferred per 2026-04-25 audit log line 804.
  5. Cand 11 (ip-pop-architectural-hierarchy) — promoted to live. Targeted raw check of PoP "p. 79 Heidegger malaise" passage flagged in Counterpressure as a tightening that would not block but would strengthen the evidence chain.

Duplicate-merge decisions

  • Cand 10 + Cand 12 merged into ineinander-universalizes-institution per kickoff specification. The two seed candidates were near-twins: (10) the Ineinander-universalizes-institution thesis and (12) the symbolic-matrix-as-instance-of-Ineinander corollary. The merged entry treats the universalization thesis as the main claim and the symbolic-matrix-as-instance as a sub-thesis within the same claim. No alternate-slug cross-references inside Status History; the merge happened pre-creation, not post-creation, so there are no orphan slugs to preserve.

No other duplicate-merge decisions were made this run. The 5 live entries cover distinct theses (joint-operation, negative-evidence corpus, literary-language gap, Ineinander universalization, architectural hierarchy); the 5 candidate entries cover distinct philological / attribution / corrective theses with distinct evidence-chain dependencies.

False-friend cautions added

  • Heinbokel's déformation cohérente in medical-phenomenology register — added as Open Questions item + False-Friend Caution subsection on [[coherent-deformation]] (Pass 4 outcome, recorded under that page's Open Questions / False-Friend Caution subsection). The caution is opening-the-question framed: Heinbokel 2024 is in raw/ but unindexed; if a future ingest reveals genuine secondary-literature contestation, the caution can be promoted to a corrective claim entry. This is exactly the disposition the kickoff specified for Cand 8.

Supported-promotion confirmations (R8)

0 supported promotions filed for confirmation. Per R8: any supported-gate consideration in Claude Code mode halts and writes the question to wiki/.audit/phase-8-supported-confirmations-{date}.md for follow-up. None of the 5 live candidates was assessed as plainly meeting the 5-test gate on first walk. Conservative first-run target was 0 supported promotions; that target was met without R8 halts.

The 5-test gate criterion 5 ("the confidence rating survives restatement of the claim against its strongest counter-position") is the binding criterion in each case:

  • science-secrete-stiftung-chiasm — confidence: medium. Restated against the chiasm-centric secondary reading (Carbone, Fóti, Johnson, Kaushik), medium does survive but high would not; staying at live with medium is right.
  • ingested-corpus-four-element-gap — confidence: medium. Restated against the corpus-boundary objection (the claim could be falsified by Saint Aubert 2006 or Faul ingest), medium survives. Promotion to supported would require either the missing ingests or pre-authorization to treat the claim as restricted-corpus-only.
  • kaushik-stiftung-literary-frame — confidence: medium. Restated against the "Kaushik does engage painting in 2019 ch. 2" reading, medium survives because the gap is primary-framing not topical-absence; high would over-claim.
  • ineinander-universalizes-institution — confidence: medium. Restated against the Saint Aubert surrection and Kaushik matrix-primacy rivals, medium survives. Promotion to supported would need either ruling on the rivals or pre-authorization that the Counterpressure documentation is sufficient.
  • ip-pop-architectural-hierarchy — confidence: medium. Restated against the dominant late-MP-moves-away-from-institution reading, medium survives. The PoP p. 79 raw-check would tighten but does not block live status.

In each case the right disposition is live with medium confidence; promotion to supported is a follow-up message decision per R8.

Cite-back (Step 8.9 / migration kickoff Pass 5)

After the run's claim entries are committed, walk the Wiki homes lists in each promoted entry and update the cited concept pages with provisional live-claim citations. Pass 5 is performed in this same audit run, separately committed.

Wiki-homes inventory for the 5 live promotions:

Distinct pages to touch: science-secrete, stiftung, chiasm, coherent-deformation, indirect-ontology, symbolic-matrix, ineinander, institution (8 pages total).

Per kickoff Pass 5 specification: live claims get provisional citation "see claims#claim-slug (live claim)"; supported claims get normal citation without provisional framing. All 5 of this run's promotions are live, so all citations are provisional. Each page also receives an updated bump per General Rule 3 (substantive interpretive content added — claim citations are interpretive synthesis, not metadata).

Step 8.8 — Post-Phase-8 confidence addendum

Per CLAUDE.md v0c §Workflow: Audit "Note on phase order": Phase 7 numerically precedes Phase 8 but Phase 7's confidence-recalibration logic depends on Phase 8 outputs when synthetic-layer work occurred. The 2026-04-25 audit completed Phase 7 before Phase 8 existed; this addendum addresses pages whose confidence may have changed because of synthetic-layer assembly.

Pages potentially affected by synthetic-layer work:

  • science-secrete — Pass 3b upgraded to v0c sections (confidence already recalibrated 2026-04-27). The cite-back to claims#science-secrete-stiftung-chiasm (live) and claims#ingested-corpus-four-element-gap (live) does not require further confidence change: the page's claims are now backed by synthetic-layer entries that are themselves at medium confidence with documented Counterpressure. The page-level confidence: high rating remains adequate because the page does not over-claim — it presents the synthesis as articulated rather than settled.
  • coherent-deformation — confidence: high. Adding the Heinbokel false-friend caution (this run) and the cite-back to two live claims does not change the page's epistemic profile; the page accurately represents its sources and now flags an external false-friend caution. No confidence change.
  • stiftung — Pass 3a-created page (confidence: high, recalibrated 2026-04-27). The cite-back to four live claims (corpus-gap, kaushik-literary-frame, ineinander-universalization, ip-pop-architectural-hierarchy) reinforces the page's positioning without changing its epistemic profile. No confidence change.
  • chiasm — confidence: high. Cite-back to two live claims (joint-operation, corpus-gap) is consonant with the page's existing Stakes and Problem-Space sections; no confidence change.
  • indirect-ontology — confidence: medium, epistemic_status: contested (Pass 3a). The cite-back to three live claims further supports the contested-but-articulated framing; no confidence change.
  • symbolic-matrix — confidence: high. Cite-back to two live claims (kaushik-literary-frame, ineinander-universalization) consolidates the kickoff's Cand 9 + Cand 10+12 framings into the synthetic layer; no confidence change.
  • ineinander — confidence: high. Cite-back to two live claims (ineinander-universalization, ip-pop-architectural-hierarchy) consolidates the universalization thesis; no confidence change.
  • institution — confidence: high. Cite-back to two live claims (ineinander-universalization, ip-pop-architectural-hierarchy) reinforces the institution-not-abandoned reading; no confidence change.

Net confidence-addendum verdict: no concept-page confidence changes resulting from Phase 8. The synthetic-layer work makes existing interpretive commitments explicit; it does not introduce new uncertainty or new resolution. This is the expected pattern when Phase 8 runs against a corpus that has just been Pass-3-upgraded — the synthetic layer crystallizes work already done at the page layer, rather than disturbing it.

Outputs produced this run

  • wiki/claims.md — populated (replaced the stub) with 5 live + 5 candidate entries plus the False-friend cautions section.
  • wiki/.audit/synthetic-layer-2026-04-27.md — this report.
  • wiki/concepts/coherent-deformation.md — Heinbokel False-Friend Caution subsection added under Open Questions.
  • (Pass 5 cite-back) — 8 concept pages touched with provisional live-claim citations.
  • wiki/index.md — Claims summary table populated; Claim entries count updated.
  • wiki/overview.md — Recent Activity entry; Current Stats Claim entries.
  • wiki/log.md — Phase 8 entry appended.

Stop conditions encountered

None. No untraceable seed candidate forced retreat to verbal/informal support (Rule 17 hard rule held); no targeted-raw-check exposed a discrepancy (because the in-run targeted raw checks were deferred to candidate Counterpressure rather than blocking live promotions); cap not exceeded (5 live, 0 supported, well within 10–15 / 3–5 budget).

Next-audit-run handoff

For the next audit run that includes Phase 8:

  • Walk the 5 candidate-only entries' evidence chains if the missing artifacts have been ingested (Faul, Valéry Œuvres II, Heinbokel, Malraux Voices of Silence) or if targeted raw checks have been performed (V&I Fr 320–321, PoP p. 79).
  • Walk the deferred wiki/motifs.md BRIDGE entries (item 3 of snapshot triage Phase 8 walk-order remaining) for 1–3 additional structural-parallel claim entries.
  • Walk the deferred extraction-note Diagnostics + Recurring Motifs sample (snapshot triage item 1: E&M, V&I, I&P, PoP, Beith, Knight) for any additional candidates beyond the kickoff seed list.
  • Subsequent Phase 8 runs update claims.md rather than create it. Per AUDIT_PLAN v1.4 §Subsequent runs: "Read existing entries before walking new artifacts; look for claims whose status should change (e.g., live → supported on additional evidence; supported → contested if a later source materially complicates the claim)."

Bottom line

The first Phase 8 run produces a 10-entry claims register: 5 live promotions covering the joint-operation thesis, the negative-evidence corpus claim, the Kaushik literary-language gap, the Ineinander universalization (with merged corollary), and the I&P → PoP architectural hierarchy; and 5 candidate-only entries with explicit evidence-chain preconditions. One false-friend caution was added (Heinbokel on [[coherent-deformation]]). Zero supported promotions: R8 was honored without filing follow-up confirmations because no candidate plainly met the 5-test gate. Pass 5 cite-back follows in this same run, separately committed.

The synthetic interpretive layer is now articulated as evidence-bearing claim entries rather than implicit page commitments. The wiki's contribution to Paper A's preparation is legible as more than aggregation: the 5 live claims are exactly the cross-cutting reading the secondary corpus does not compose.